Single-cell RNA Sequencing Technology Revealed the Pivotal Role of Fibroblast Heterogeneity in Ang II-Induced Abdominal Aortic Aneurysms

Aim: The mechanism of abdominal aortic aneurysm (AAA) has not been fully elucidated. In this study, we aimed to map the cellular heterogeneity, molecular alteration, and functional transformation of angiotensin (Ang) II-induced AAA in mice based on single-cell RNA sequencing (sc-RNA seq) technology. Method: Single-cell RNA sequencing was performed on suprarenal abdominal aorta from male APOE-/-C57BL/6 mice of Ang II-induced AAA and shame models. Immunohistochemistry was used to determine the pathophysiological characteristics of AAA, and sc-RNA seq was used to determine the heterogeneity and phenotypic transformation of all cell types. A single-cell trajectory was performed to predict the differentiation of broblasts. Finally ligand–receptor analysis was used to evaluate intercellular communication between broblasts and smooth muscle cells. Results: More than 27,000 cells were isolated and 25 clusters representing 8 types of cells were identied, including broblasts, macrophages, endothelial cells, smooth muscle cells, T lymphocytes, B lymphocytes, granulocytes, and natural killer cells. During AAA progression, the function and phenotype of different type cells altered separately. The pro-inammatory function of inammatory cells was enhanced. The proliferation phenotype degreased while pro-inammatory, regeneration and damage-related phenotypes increased in endothelial cells. Smooth muscle cells also transformed from contractile to secretory phenotype. The alterations of broblasts were the most conspicuous according sub-group clustering analysis. Single-cell trajectory revealed the critical reprogramming genes of broblasts mainly enriched in regulation of immune system. Finally, the ligand–receptor analysis conrmed that increases in secondary collagen synthesis led by broblasts were one of the most prominent characteristics of Ang II-induced AAA. Conclusion: Our study revealed the cellular heterogeneity of Ang II-induced AAA. Fibroblasts may play a central role in Ang II-induced AAA progression according multiple biological functions including immune regulation and extracellular matrix metabolic balance. Our study may provide us with a different perspective on the etiology


Background
Abdominal aortic aneurysm (AAA) is a common fatal cardio great vessel disease worldwide, characterized by a degree of local dilation ≥ 50% compared with that of adjacent normal tissue [1].
Although research on AAA has been ongoing for decades, it's etiology remains controversial [2,3].
Currently, there are several hot topics surrounding the pathogenesis of AAA. 1) The connection between AAA and atherosclerosis: Conventional view was that AAA was just a result/manifestation of atherosclerosis. However, researchers found that AAA seems to be independent of atherosclerosis, from the perspective of clinical, genetic, and biochemical studies in recent years [3]. It is unquestionable that both AAA and atherosclerosis share some common pathophysiological characteristics, such as chronic vascular in ammation, vascular remodeling, and involvement of the renin-angiotensin-aldosterone aneurysm [14]. A single model can only re ect one aspect of the AAA. Previous studies have explored the cellular heterogeneity of elastase-induced AAA and CaCl 2 -impregnated AAA models [11,15]. Mapping the cellular heterogeneity of Ang II-induced AAA in mice will complete the research on the mechanism of the disease. Transcriptome sequencing of Ang-II-induced AAA at the single-cell level may become one piece of the puzzle in the on the mechanism researches of AAA.
In this study, we constructed Ang II-induced AAA models for 10x Genomics sc-RNA seq to analyze the cellular heterogeneity and molecular alterations of the disease which may be complement genetic studies of aortic disease. Our research might provide different perspectives for understanding AAA progression.

Animals and rearing condition
A total of 30 male Apoe-/-C57Bl/6J mice (10 weeks) (purchased from GemPharmatech Co., Ltd, Soochow, China) were used for the experiments. All mice were raised in the speci c pathogen free (SPF) laboratory of Zhejiang Academy of Medical Sciences and were randomly divided into two equal groups (n = 15), namely, the control (CTR) group and the AAA group. Mice were housed communally with a standard 12 h dark cycle and fed ad libitum. The room temperature was controlled at 20-25 ℃, and the humidity was controlled at 50-60%. All our operations were in line with"Guiding Principles in the Care and Use of Animals" (China). Ethics approval was obtained from laboratory of Zhejiang Academy of Medical Sciences (No. 20200271) prior to the start of the study.

Construction of AAA models
All of the mice were used for model construction after 2 weeks of adaptation and pre-feeding. To generate the model, the AAA group was administered Ang II solution (1000 ng/kg/min, Sigma Chemical Co., St Louis, MO, USA) via Alzet osmotic minipumps 2004 (Durect Corporation, Cupertino, USA), while the CTR group was administered normal saline for 4 weeks. Brie y, after general anesthesia of the mice with iso urane (4%, 400 ml/min, Abbot, AbbVieLtd., United Kingdom), an incision was made on the back of the mouse to create a cavity. The osmotic minipump was placed into the cavity, and the incision was sutured. Body weight was measured at day 0 (before implanting the osmotic minipump), day 1, and 1, 2, 3, and 4 weekends after the implantation of the minipumps. The daily physiological state of the mice was determined. Once a mouse died, an autopsy was performed to determine the cause of death.

Ultrasonic inspection and measurement
The diameter of the abdominal aorta and cardiac function were measured using the Visual Sonics Vevo 770 (Visual Sonics, Toronto, Canada) ultrasound imaging system, before the mice were sacri ced. Bmode imaging was performed to measure the diameter of the abdominal aorta while M-mode imaging was performed to measure cardiac function. To ensure reproducibility and accuracy of the experiment, all operations were performed by the professional operator who was blinded to the grouping beforehand.

Collection of tissue samples and serum
The mice were anesthetized with iso urane (4%, 400 ml/min), and the chest was cut open. The hearts of the mice were exposed, and blood was drawn from the ventriculus dexter. Thereafter, 10-15 ml of normal saline was injected into the ventriculus sinister to ush the residual blood in the aorta. The suprarenal abdominal aorta segment was separated under a stereoscopic microscope and placed into a tissue preservation uid (Shbio, Shanghai, China) for cell digestion and sc-RNA seq analysis (n = 8 in each group). The extracted blood was placed in coagulation tubes at 20°C for 2 h and then centrifuged at 3000 rpm for 10 min. The supernatant was collected to obtain the serum for subsequent biochemical index determination. Additionally, the abdominal aorta tissue of mice from each group (n=4) was xed with 5% paraformaldehyde for immunohistochemical (IHC) and immuno uorescence (IF) staining.

IHC, IF and special stain
The xed abdominal aorta tissues were embedded in para n and cut into 2-µm-thick serial sections for IHC staining. The dyeing steps were as follows: 1) dewaxing and hydration: the sections were incubated in xylene for 10 min at 60°C (twice), the sections were then placed in absolute ethyl alcohol for 5 min (twice), followed by placement in 95% alcohol, 80% alcohol, and 70% alcohol for 2 min each, and nally, the sections were washed with distilled water (three times). 2) Epitope retrieval: The sections were immersed in 0.01 M sodium citrate buffer solution, heated to boil, and then heated for 6 min (four times Biosciences, 1:50) at 4°C overnight. After washing with PBS four times, the sections were incubated with biotin-coupled secondary antibodies at 37°C for 30 min. 5) Coloration: sections were incubated with horseradish peroxidase (HRP) for 20 min at room temperature. After washing with PBS three times, the sections were stained with 0.05% diaminobenzidine (DAB) for 5 min. The sections were then washed with PBS for 3 min (3 times). 6) Counterstaining: The sections were stained with hematoxylin solution for 5 s and washed with PBS until the water was not discolored. 7) Dehydration, clearing, and sealing: The sections were successively incubated with 50% alcohol, 70% alcohol, 95% alcohol, and absolute ethyl alcohol (twice) for 1.5 min each. After incubation with xylene for 1 min (twice), the sections were sealed with a moderate amount of neutral gum.
The dyeing steps of IF were slightly different with IHC and as follow: After epitope retrieval, the sections were incubated with 20 µl 5% FBS in PBS at 20°C for 30 min. Then the sections were co-incubated with primary antibody against vimentin [Vim, (60330-1-Ig, Proteintech, 1:50)] and one of the follow antibody: The steps of hematoxylin-eosin (HE), Masson, and Verhoeffs Van Giesof (EVG) dyeing were similar to the steps mentioned above and were adjusted according to the manufacturers' instructions. The degree of degradation of elastic bers was de ned based on previous studies: 1 <25%, 2 ≤25%-≤50%, 3 <50%-≤75%, and 4 >75% [11].

Single cell preparation and RNA sequencing
The abdominal aorta tissue of the mice was washed with 4°C normal saline solution containing 3% FBS (Thermo Fisher Scienti c, Inc.) to remove attached blood clots. The tissue was cut into pieces as small as possible and dissociated into individual cells according to the protocol of the 10× Genomics platform and previous research [11,16]. Thereafter, arterial tissue fragments were incubated in a speci c enzyme solution containing 125 U/ml collagenase type XI (C7657, Sigma, USA), 450 U/ml collagenase type I (C0130, Sigma), 60 U/ml hyaluronidase type I-S (H3506, Sigma), and 60 U/ml DNase I (DNASE10, Sigma) for 1 h at 37°C. The cell suspension was washed twice with normal saline containing 3% FBS, and cell debris was removed through a 70 µl lter. We used a chemiluminescent immunoassay to detect the cell activity. The cells were subjected to sc-RNA seq only when the activity was greater than 85%. Finally, the cell concentration was adjusted to 700/µl.
The whole work ow of sc-RNA seq was based on the protocols of Standard 10x Chromium Single Cell 3' Solution v2 System (10X Genomics Gemcode Technology). The prepared single-cell suspension, 10×barcode gel magnetic beads, and oil droplets were added to different channels of chromium chip B, and the gel beads-in-emulsions (GEMs) were formed through the micro uidic cross-ow system. Each GEM acted as a separate reaction system. The GEMs owed into the reservoir and were collected. The gel beads were dissolved to release the barcode sequence, which was reverse-transcribed to the cDNA fragment, after which the sample was tagged. After the liquid reservoir was destroyed, PCR ampli cation was performed using cDNA as a template. All GEM products were mixed to build a standard sequencing library. The double-ended sequencing mode of the Illumina sequencing platform was used to carry out high-throughput sequencing of the constructed libraries.

Sequencing data processing and analysis
The 10× Genomics Cell Ranger software was used to perform data quality statistical tests on the original data and compare with the reference genomes. Read-matched genomes were divided into exons, introns, and intergenic regions (at least 50% of the bases were matched to exons, introns, and intergenic regions), and when reads were simultaneously matched to an exon and other non-exons, these reads were classi ed as exon reads in priority. We further analyzed the exon reads combined with the annotation.
The reads were considered to be transcriptome reads when matched to the exon of the transcriptome and when bases were arranged in the same direction. If transcriptome reads were mapped to only one gene, they were unique-mapped, and only unique-mapped reads were used for unique molecular identi er (UMI) counting.
Further cell ltration, standardization, cell subpopulation classi cation, differentially expressed gene (DEP) analysis of each subpopulation, and marker gene screening were performed using the R package of Seurat v3.0. We ltered the cells based on the number of genes expressed. The speci c screening criteria for high-quality cells were as follows: 1) The number of genes identi ed in a single cell was at least 500; 2) the proportion of mitochondrial gene expression in single cells was less than 25%; 3) Multicellularity was removed using the DoubleFinder package. Data were then log-normalized for subsequent analyses by normalization.
The classi cation and analysis process of cell subpopulations was as follows: 1) After dimensionality reduction, principal component analysis (PCA) was used to process the normalized expression values. The rst ten principal components were selected from the PCA analysis results for subsequent clustering and clustering analysis. 2) The signi cant principal components were used to construct the k-nearest neighbor clustering diagram based on Euclidean distance. 3) The Jaccard similarity coe cient was used to optimize the weight value of the intercellular distance. 4) A clustering algorithm based on the shared nearest neighbor module optimization was used to identify the cell clusters.
DEPs in different cell clusters were analyzed using the bimod likelihood ratio statistical test to screen for upregulated genes in different cell populations. The screening criteria for upregulated genes were as follows: 1) The genes were expressed in more than 10% of the AAA or CTR subpopulations; 2) a P-value ≤ 0.01; 3) a fold change of upregulated genes ≥ 1.5.

Gene ontology and Kyoto Encyclopedia of Genes and Genomes analyses
In organisms, different genes coordinate their biological functions with each other. Pathway-based analysis is helpful for further understanding the biological functions of genes. Functional enrichment analysis was performed using the gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) software online (https://david.ncifcrf.gov/home.jsp ). The gene names of the DEGs were listed in an Excel table and uploaded to the online software. The analysis was carried out according to the website's instructions. Enrichment analysis was predominantly used to determine the enrichment degree of differential genes in the GO terms. Hypergeometric tests were used to evaluate the signi cance of GO terms. The identi ed pathways that were signi cantly enriched in DEPs were compared to the entire genome. Statistical differences were considered to be signi cant when P-value ≤ 0.05.

Inference of broblast state using trajectory analysis
Pseudotime trajectory analysis of single-cell transcriptomes was conducted using Monocle 2. The subsequent analysis steps refer to the process of the o cial website (http://cole-trapnell-lab.github.io/monocle-release/docs/#getting-started-with-monocle). We rst stored data in a CellDataSet object. The ordering work ow for constructing a single cell trajectory was as follows: 1) Choose the genes that de ned progress; 2) reduce dimensions; 3) pseudotime analysis: UMIs were modeled with the negative binomial. Genes that determined the fate of differentiation were selected and processed by metascape software online (https://metascape.org/gp/index.html#/main/step1 ) to rther reveal biological functions.

Analysis of intercellular interactions between different cell types
In the following analysis, we used ligand-receptor interactions to evaluate the interactions between different cell types according to the CellPhoneDB database (https://www.cellphonedb.org/documentation). The strength of ligand-receptor interaction between cell types (the interaction score) was de ned as the product of the average expression of ligands in cell type A and the average expression of receptors in cell type B. Additionally, we constructed the background distribution of interaction scores based on the idea of perturbation and obtained P-values according to the real interaction scores. Ligand-receptor pairs with P values ≤ 0.05 were selected as targets.

Statistical analysis
Data analysis and gures making were performed by using GraphPad Prism software (version 5.0; GraphPad Software, San Diego, CA, USA). When the data conformed to normal distribution and homogeneity of variance, Student's t test was used to evaluate the statistical signi cance, and the results were presented as mean ± standard deviation. Alternatively, Kruskal-Wallis tests were performed on nonnormally distributed counting variables. These results were presented as median (quarterback spacing). The Kaplan-Meier test was used to evaluate the difference in survival rate. The relative positive area was calculated by Image J. Differences were considered statistically signi cant at P ≤ 0.05. The whole work ow was shown in supplement gure.1.

Construction of AAA models and associated physiopathologic characteristics
The weights of all mice were dynamically monitored for 4 weeks after the implantation of minipumps. There was no difference in body weight between the two groups with an increase in feeding time (Supplementary Fig. 2A). One mouse died of AAA rupture on day 26 in the AAA group, while all other mice survived. There was no signi cant difference in the survival rate between the two groups (P = 0.373, Supplementary Fig. 2B). No signi cant differences were found in creatine kinase (CK), creatine kinase MB (CKMB), lactic dehydrogenase (LDH), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and total cholesterol (TC) between the two groups. Interestingly, the levels of blood glucose (GLU) and total triglyceride (TG) were lower in the AAA group, which may be due to a reduction in food intake ( Supplementary Fig. 2C).
Ultrasound examination revealed that the intraluminal diameter and intima thickness of the AAA group tended to increase not accompanied by alternation of cardiac function after continuous pumping of Ang II for 28 days ( Fig. 1A-1C, Table.1). HE staining results revealed that the basic structure of the artery had been destroyed, and in ammatory cell in ltration was mainly located in the adventitia and media of the artery wall in the AAA group. Masson staining results revealed that SMCs showed a disorderly arrangement accompanied by secondary collagen hyperplasia in the AAA group. EVG staining revealed that the break of elastin layer in the AAA group. Small new elastic bers were observed in the crevasse (Fig. 1D). Semi-quantitative analysis indicated that there was a tendency for elastin degradation to increase (Fig. 1E). The expression of CD68 and MCP-1 increased in the AAA group, which revealed in ltration of macrophages/monocytes in the arterial wall tissue. Moreover, the expression of cytokines, including TNF-α and IL-1β, increased in the arterial wall tissue (

Single cell sequencing identi ed a total of 25 cell clusters representing eight types of cells
The overall quality parameters of the sc-RNA sequencing data were as follows: A total of 13,779 and 14,086 cells were obtained in the CTR and AAA groups, respectively. The median number of genes per cell was 2734 and 1818 in the CTR and AAA groups, respectively ( Supplementary Fig. 3A). Cell activity was an important factor that affected the results. After the removal of low-activity and dead cells, the distribution of the number of genes detected (nFeature_RNA), the total quantity distribution of UMI detected (nCount_RNA), and the percentage of mitochondrial gene expression (Percent.mito) in single cells were obtained ( Supplementary Fig. 3B). Cells with mRNA >500 and a proportion of <25% of mitochondrial genes met the standard and were used for subsequent data analysis.
The non-linear dimensionality reduction method of t-distributed stochastic neighbor embedding (t-SNE) was used to visualize the cell subpopulation classi cation results. A total of 25 cell clusters were isolated from the AAA and CTR groups. In the development of AAA, the cell types, the number of clusters, and the proportion of cells in each group were altered ( Fig. 2A).
T and B cells are important components of the immune cells. In this experiment, fewer T and B cells were identi ed in the AAA group (Fig. 3A). During the development of AAA, the expression of Cd3 and Cd8 increased signi cantly, while Cd4 expression was similar in both groups (Fig. 3G). Bioinformatic analysis further revealed the functional alternations of T lymphocytes. Ribosome-related components, translation, T cell differentiation and activation, protein binding, and the structural constituent of ribosomes accounted for the majority of the GO enrichment terms. During the formation of AAA, the protein synthesis of T lymphocytes increased, and the differentiation and activation of cells was vibrant ( Supplementary Fig. 4A). KEGG pathway analysis suggested that protein synthesis (ribosome) and T cell activation and differentiation were the most critical mechanism alterations ( Supplementary Fig. 4B).
After activation, B lymphocytes can differentiate into plasma cells. In addition to synthesizing and secreting various immunoglobulins, plasma cells also highly express Cd38 and Cd27, while there is a low expression of Cd5, which is consistent with the current results (Fig. 3H). Further bioinformatic analysis suggested that ribosome-related components, immune system process, translation, cytoplasmic translation, mRNA processing, and protein binding accounted for the majority of the GO enrichment terms ( Supplementary Fig. 4C). KEGG pathway analysis suggested that the B cell receptor signaling pathway, protein synthesis (ribosome) pathway, and NF-κB signaling pathway were critical pathways in the AAA process. Additionally, the activation of B cells also affected the differentiation and function of T lymphocytes ( Supplementary Fig. 4D).

Transformation in the function and phenotype of SMCs and ECs during the formation of AAA
SMCs and ECs are important components of the artery wall tissue. Corresponding functional and phenotypic changes in SMCs and ECs also occurred during the disease process of AAA. The numbers of EC1 and EC3 generally decreased, while that of EC2 increased during AAA formation, and numbers of both SMC1 and SMC2 increased (Fig. 4A). The top 10 marker genes of ECs were identi ed for each cluster relative to all other clusters. The expression patterns of marker genes in EC1 and EC3 were similar, which was different from that of EC2 (Fig. 4B). Phlogosis and damage-related genes, including vascular cell adhesion molecule-1 (Vcam1), von Willebrand factor (Vwf), intercellular adhesion molecule 1 (Icam1), endothelin 1 (Edn1), Serpine1, and prostaglandin I2 synthase (Ptgis) [19][20][21][22], were highly expressed in EC2. Proliferation and regeneration-related genes, including endoglin (Eng), kinase insert domain receptor (Kdr), and chromodomain helicase DNA-binding protein 5 (Chd5) [23][24][25], were highly expressed in EC1 and 3 (Fig. 4C), which was further con rmed by the violin plot (Fig. 4D).
Similar phenotypic and functional changes were observed in the SMC. The expression of secretory SMC marker genes [secreted phosphoprotein 1 (Spp1), matrix Gla protein (Mgp), epiregulin (Ereg), and elastin (Eln)] increased in SMC1 and SMC2, while the expression of contractile SMC marker genes [Acta2, Tagln, caldesmon 1 (Cald1), and Myh11] remained unchanged during AAA formation (Fig. 4E), which implied a transition of SMCs from contractile to secretory during the formation of AAA.

The cell heterogeneity and DEG expression of Fbs in AAA
Fibroblasts were identi ed into most clusters in our study. During the AAA process, the amount of Fb1, Fb4, Fb5, Fb9, and Fb10 decreased while Fb2, Fb3, Fb6, and Fb8 increased, Fb7 was basically unchanged (Fig. 5A). The top 10 marker genes were identi ed for each cluster relative to all the other clusters.
Clusters with the same trends in amount had similar gene expression mode (Fig. 5B). Further screening of DEGs was necessary. A total of 411 upregulated genes and 605 downregulated genes were identi ed (Fig. 5C). GO enrichment analysis revealed that Fbs were predominantly involved in the binding of collagen and the synthesis and degradation of ECM (Fig. 5D). KEGG pathway analysis revealed that oxidative phosphorylation, ECM-receptor interaction, and PI3K-Akt signaling pathway may be the pivotal signaling pathways that participate in the formation of AAA (Fig. 5E).

Single cell trajectory analysis predicted cell differentiation of Fbs in AAA
The process of phenotypic and functional alternations in the cells was gradual. Therefore, it was not clear which pattern of Fbs was responsible for AAA initiation. The pseudo-time of Fbs with gene expression pro les of different clusters was reconstituted (Fig. 6A). Gene expression could be roughly divided into seven stages according to the different nodes (Fig. 6B). The typical direction of Fb differentiation was the transformation from stage 1 to stage 4. The t-SNE map also re ected the direction of cell differentiation; dark-colored cells continued to transform into light-colored cells (Fig. 6C). We identi ed that Fb5 was the starting point of differentiation and ultimately differentiated into Fb2 and Fb6. The heatmap displays the alternation tendencies of the top 50 critical genes. Overall, we found three different patterns of genetic changes, namely, Cluster 1: The expression of genes rst increased and then decreased with the development of AAA, including that of ribosomal protein L37a (Rpl37a), ribosomal protein S20 (Rps20), and ribosomal protein S27 (Rps27); Cluster 2: The expression of genes decreased with the development of AAA, including that of myelin and lymphocyte protein, T cell differentiation protein (Mal) and noncompact myelin-associated protein (Ncmap); Cluster 3: The expression of genes increased with the development of AAA, including that of complement component factor h (Cfh), biglycan (Bgn), and v-maf musculoaponeurotic brosarcoma oncogene family, protein B (Mafb) (Fig. 6D). The top six genes displaying the most distinct expression change were considered to have determined the outcome of Fb differentiation. The relative expression levels of the top six genes were shown over pseudotime by Monocle2 in cluster mode (Fig. 6E) and state mode (Fig. 6F).
As we selected the top 50 genes which determined the direction of Fbs differentiation, further bioinformatics studies were necessary. The overlap of the 3 clusters was shown in the circus plot and the blue curves link genes that belong to the same enriched ontology term (Fig. 6G). The heatmap of GO enrichment terms shown that genes listed in cluster 3 (up expressed) mainly participated in negative regulation of immune system process, regulation of complement cascade and antigen processing and present; genes in cluster 2 (down expressed) mainly participated in myelination; genes in cluster 1 ( rst up and then down expressed) mainly participated in SRP-dependent cotranslational protein targeting to membrane, negative regulation of peptidase activity, post-translational protein phosphorylation, platelet degranulation, protein localization to membrane, response to endoplasmic reticulum stress, vesicle organization and response to metal ion (Fig. 6H). To further capture the relationships between the terms, a subset of enriched terms had been selected and rendered as a network plot (Fig. 6I). Although no common GO enrichment terms were found between the 3 clusters, it seemed that cluster 1 was associated with cluster 3 according part of the GO terms, which involved in regulation of the immune system, phosphorylation of post-translational protein and degranulation of platelets.

Analysis of marker genes revealed different subtypes of increased broblasts in function and spatial distribution
It was necessary to identify Fbs through the expression of speci c markers. Seven different terms were identi ed, namely, normal quiescent state [Vim and caveolin 1 (Cav1)] [26], active state [S100a8, broblast activation protein alpha (Fap), Acta2, platelet-derived growth factor receptor beta (Pdgfrb)] [27,28], antigen presentation [Cd74 and histocompatibility 2, class II antigen A, beta 1 (H2-Ab1)] [29], phlogosis (Il1b and Il6) [30], angiogenesis (Des) [31], extracellular matrix synthesis [platelet-derived growth factor receptor alpha (Pdgfra) and bulin 1 (Fbln1)] [32] and tissue repair [ bronectin 1 (Fn1)] [33]. The dot plot showed the expression of characteristic genes in different Fb clusters (Fig. 7A). Based on the results from single-cell trajectory analysis, Fb5 might be the starting point of differentiation (Fig. 6C) and was selected as the control cluster. Semi-quantitative analysis showed the difference in the expression of related functional genes between Fb5 and increased Fbs in the expression of related functional genes (Fig. 7B). The expression of marker genes in increased Fbs was compared with that of Fb5, and the values of fold change (FC) were obtained. The increased Fbs were rede ned according to the FC values (Fig. 7C). The criteria were as follows: 0 (0.5<FC value<2); + (2≤FC value<4); ++ (4≤FC value<8); +++ (8≤FC value); -(0.25<FC value≤0.5); --(0.125<FC value≤0.25); ---(FC value≤0.125). Fb2 highly expressed Cd74 and H2-Ab1 and was de ned as an antigen presenting Fb (apFb). Fb6 highly expressed in Fn1 and was de ned as tissue repair Fb (trFb). Fb8 highly expressed Des and was de ned as vascular Fb (vFb). Fb3 only highly expressed in activation-related markers and was de ned as activation Fb (acFb). Immuno uorescence con rmed the presence of 4 Fbs in AAA tissue (Fig. 7D), Fb2 co-expressed Vim and Cd74 and can be observed in the whole artery layer of AAA tissue; Fb3 highly co-expressed Fbln1 and Vim, and mainly distributed in the intima of arteries; Fb6 highly co-expressed Fn1 and Vim, which was mainly observed around thrombus; Fb8 high co-expressed Des and Vim. As the smallest subgroup, it was hardly been discovered for its small amount.

The ligand-receptor interaction analysis revealed the procollagen synthesis effect of Fbs on SMCs
Cell-cell communication was investigated by ligand-receptor analysis according to CellPhoneDB. The results indicated that the intensity of the interaction between different types of cells varied greatly. Meanwhile, the Fbs had the closest connection with the SMCs (Fig. 8A). We further processed the ligandreceptor analysis between SMCs and increased Fbs (Fb2, Fb3, Fb6, and Fb8) during AAA formation. Fb3 and Fb6 were found to have more connections with SMCs than Fb2 and Fb8 (Fig. 8B). Four distinct Fb clusters in uenced the function and phenotype of SMC through multiple ligand-receptor pairs (Fig. 8C-F). Different Fb/SMC pairs showed speci c ligand-receptor characteristics. We found that CD74macrophage migration inhibitory factor (Mif), Cd74-COPI coat complex subunit alpha (Copa), and CD74amyloid beta precursor protein (App) displayed important associations between Fb2 and SMCs, con rming the antigen-presenting effect of Fb2. The Fn1-a5b1 complex, Fbn1-a5b1 complex, Fn1-a8b1 complex, Fn1-aVb1 complex, and Fn1-aVb5 complex were critical links between Fb6 and SMC, which demonstrated that Fb6 played an important role in tissue repair during disease progression. Alternatively, increased collagen synthesis was the most prominent feature of the selected Fbs, except Fb8. Additionally, the expression of Spp1-a9b1 complex and broblast growth factor receptors (Fgfr1)neural cell adhesion molecule 1 (Ncam1) commonly increased signi cantly while neuropilin-1 (Nrp1)vascular endothelial growth factor B (Vegfb) and Pdgfr complex-platelet-derived growth factor D (Pdgfd) were decreased in 4 Fbs.

Discussion
The occurrence of vascular-associated diseases is often accompanied by the functional and phenotypic transformation of different types of cells. Sc-RNA seq provides critical support for elucidating the mechanism of disease by indicating cellular heterogeneity. In this experiment, we processed sc-RNA seq data of >20,000 individual cells from the abdominal aorta tissue of the upper renal artery that was exposed to either Ang II or to normal saline. Finally, 25 clusters and eight different cell lines were identi ed. Based on the marker genes expressed in each cell cluster, we explained the transformations in cellular function and phenotype in Ang II-induced AAA. Further bioinformatics analysis of DEGs and differentiation trajectory revealed that Fb played an important role in the synthesis and degradation of ECM and the regulation of immune system. Finally, the ligand-receptor analysis between increased Fbs and SMCs was performed and found that increased collagen synthesis was an important feature of Ang II-induced AAA. Overall, our experiment mapped the cellular heterogeneity of Ang II-induced AAA.
Fibroblasts transformed into different phenotypes during the course of disease and participated in regulation of immune system and the metabolic homeostasis in ECM. Its heterogeneity played a critical role in pathophysiological process of Ang II-induced AAA.
Comparisons with previous studies would help us better understand our results. Hadi et al. also investigated the cellular heterogeneity of arterial tissue in Ang II-induced AAA. They found that Fbs made up the largest population in the tissue of AAA which was consisted with our results [34]. It suggested that Fb may play a critical role in the pathophysiological process of Ang II-induced AAA, however the potential pathogenic mechanism had not been elucidated in Hadi's research. In the other hand, Zhao et al. found that macrophages were the largest population, and immune cells accounted for more than 60% of the total number of cells in elastase-induced AAA group. Additionally, natural killer cells, erythrocytes, and neurocytes were exclusive cell types in elastase-induced AAA models, whereas granulocytes were only found in Ang II-induced AAA [11]. The reasons for these differences may predominantly be the unequal levels of predisposing factors present within the two models.
The amount of macrophages occupied an absolutely important position in immune cells. In fact, most macrophages in the aorta were derived from circulating monocytes [35]. After colonization in tissues, macrophages perform a variety of functions, including secretion of proin ammatory factors and MMPs, activating oxidative stress response, and participating in thrombosis and vascular remodeling [36]. Our research found that the expression of M2 macrophage marker genes (mainly in Mo/Mø3, Mo/Mø5, and Mo/Mø6) decreased while M1-type macrophage marker genes, however, did not appear to increase as the AAA progressed. Previous studies had con rmed that Ang II induces the differentiation of macrophages into M1 type in AAA [37,38]. We speculated that the process of mRNA translation into polypeptides was regulated by small molecules, such as miRNA, which resulted in the increased synthesis of the marker protein. Alternations in the microenvironment were crucial for macrophage polarization, as studies have con rmed that M1-like phenotype macrophages can dedifferentiate and switch to M2-like macrophages when the microenvironment was altered [39]. M2-type macrophages play an immunoregulatory role in the formation of AAA and our results con rmed that M2a and M2c were the main M2-type macrophages in the arterial tissue, which was not been reported before. Different subtypes of M2-type macrophages played their respective physiological roles, and its role in the AAA process deserved greater attention.
Despite the small amount of identi ed T and B cells, critical phenotypic and functional changes occurred during the formation of AAA. The expression of CD3 in T lymphocytes improved in the AAA group. CD3 binds tightly to T-cell receptor to form a multisubunit complex and further induces phosphorylation of immunoreceptor tyrosine-based activation motifs, which triggers a T-cell effector response, resulting in the activation of T cells [40]. Both CD8 + and CD4 + lymphocytes have been con rmed to involved in vascular in ammation in AAA [41,42]. Our results showed that the expression of CD8 increased, while CD4 expression barely changed during the AAA process. In addition, we found that plasma cells increased in arterial tissue during AAA formation, which was con rmed by previous studies [43]. Overall, in ammatory cells tend to differentiate in a proin ammatory direction during the process of AAA.
The heterogeneity of non-immune cells is also worthy of attention. We found that the amount of EC1 and EC3, with high expression of proliferation and regeneration-related genes, decreased, while EC2 showed a high expression of in ammation and damage-related genes during the disease process in our results. The ability of ECs to produce nitric oxide (NO) is critical to adapt microenvironmental alterations. ECs contribute to the AAA process predominantly through oxidative stress mediation by impairing NO bioavailability [44,45], suggesting that dysfunction of ECs is an important characteristic of AAA. Similar functional transformations were also observed in SMC. We identi ed two clusters of SMCs in total, both of which increased during AAA formation. The number of SMCs captured in this experiment was insu cient to conclude SMCs as the most important cell ingredient in normal artery tissue. These discrepancies were likely associated with technical aspects, including the enzymatic cocktails used for single-cell release and the different approaches to data analysis [11]. With regards to cell function, we found that the expression of contractile phenotype-related genes remained largely unchanged, while the expression of secretory phenotype-related genes increased simultaneously. The literature suggests that vascular SMCs play a variety of roles in the progression of AAA. They mediate the production and degradation of ECM, transdifferentiate into macrophage-like cells, mediate in ammation, upregulate the levels of cytokines, and downregulate the expression of antioxidant genes which was consisted with our results [46]. Limited by the small number of identi ed SMCs, we only performed a simple analysis. Analysis with a larger size sample was necessary.
Fbs were the most speci c cell type identi ed; not only their number occupied an absolute advantage in the total cells, but also their extensive cellular heterogeneity and functional differences. Previous studies have con rmed that Fbs play numerous pathophysiological roles in blood vessel wall tissues, including in angiogenesis, vascular in ammation, and promoting the migration of ECs [47,48]. Analysis of DEGs provided us with a understanding of the changes in Fb function. The expression of Acta2, Tagln, and Spp1 increased, suggesting that Fbs gradually differentiated into myo broblasts during AAA formation. A study performed by Sakata demonstrated that myo broblasts participated in the vascular remodeling of in ammatory aortic aneurysm (IAA) by inducing the production of hypoxia-inducible factor 1 [49]. Cellular communication network factor 2 (Ccn2), also known as connective tissue growth factor (Ctgf), is a marker of brosis. It has been con rmed that Ctgf promotes proliferation, migration, and differentiation of Fbs [50]. Cartilage oligomeric matrix protein (Comp) is a component of ECM. Yi Fu et al found that it could act as an endogenous β-arrestin-2-selective allosteric modulator of AT1 receptor counteracting vascular injury. Its de ciency aggravated AngII-induced AAA formation [51]. Thrombospondin-1 (Thbs1) also is a matricellular protein involved in the maintenance of vascular structure and homeostasis through the regulation of cell proliferation, apoptosis, and adhesion. Liu Zhenjie et al reported that Thbs1 -/mice were resistant to aneurysm induction [52], which was different with results from another related research [53]. It suggested that the role of Thbs1 in the formation of AAA was two-sided. Complement factor D (Cfd), as a member of the alternative complement pathway, is the rate-limiting enzyme of for the formation of C3 convertase. Although the complement system has been proved to play a critical role in the development of AAA [54,55], there is no speci c study between Cfd and AAA. Due to its important role in complement systems (rate-limiting enzyme of C3 convertase), Cfd may be a potential target for the treatment of AAA.
Single-cell trajectory analysis provided further insights into the differentiation process of Fbs in AAA process. The top 50 genes that had the most substantial in uence on cell transformation were selected and roughly divided into three categories according to their expression. In brief, genes with increased expression mainly played a role in regulation of complement cascade, negative regulation of immune system process and antigen processing and present, which was consistent with the results of our DEG research. Genes with decreased expression predominantly involved in myelination, however no association between it and AAA has been reported so far. Genes with an expression that rst increased and then decreased were largely related to SRP-dependent cotranslational protein targeting to membrane and negative regulation of peptidase activity. They participated in translation, peptide metabolic processes, and ribosome assembly. The network plot revealed that immunomodulatory alternations, especially in complement system, are important features of Fb reprogramming during formation of Ang-II induced AAA.
In order to further explore the role of Fb in the AAA process, it was necessary to classify clusters of Fbs in more detail based on their functional marker genes. Previous studies on tumor provided a classi cation standard of cancer-associated broblasts (CAF) which was adopted in our research [29,31]. The 4 increased Fbs performed various functions and distributed in different areas of aneurysm tissue con rmed by immuno uorescence. Even in elastase-induced mice AAA models, broblasts also accounted for a signi cant proportion, however, its role in AAA process was underappreciate [11]. Our research con rmed that Fbs were not only involved in tissue repair, ECM metabolic homeostasis, but also in immune system regulation in aneurysm tissue. In addition, broblasts gradually transform into myo broblasts which promoted the vascular remodeling under Ang-II stimulation. Interestingly, Des and Cav1 were highly expressed in Fb8, which suggested that Fb8 was an intermediate state cell between Fbs and SMCs. Its role in the occurrence and development of AAA was still unde ned and deserved further study.
Further ligand-receptor analysis revealed the expression of molecular signals between 4 increased Fbs and SMCs. Increased collagen synthesis was the main feature of communication between Fbs and SMCs. It had been con rmed that the expression of collagen type I/III cross-links was particularly prominent in human AAA tissue, however, total collagen markers were decreased (decreased 4-hypro and 5-hylys) it was reasonable to suggest that new collagen biosynthesis was somehow defective [56]. In adition, 4 ligand-receptor pairs were common signi cantly altered in four Fbs. Spp1-a9b1, also known as osteopontin (Opn), has been con rmed to be involved in the migration of SMCs and vascular remodeling [57,58]. Ncam1 is mainly expressed in central nerve cells and can induce neurogenesis and stimulate cell-matrix adhesion and neurite outgrowth by activating Fgfr signaling [59], there are few reports of its involvement in cardiovascular disease. Nrp1-Vegfb mainly participates in the proliferation of SMCs and is associated with increased vascular diameter [60]. It also promotes the progression of in ammation and the transport of fatty acids in ECs [61]. Pdgfd was identi ed as the critical molecule in promoting vascular smooth muscle remodeling. Associated studies revealed that Pdgfd was found to be highly expressed in perivascular adipose tissue (PVAT) in AAA models of leptin-de cient mice. Pdgfd may mediate in ammation of the vascular adventitia [62]. Previous studies had con rmed that the above ligand-receptors were involved in promoting vascular remodeling and in ammatory response except Ncam1-Fgfr. However the expression of Nrp1-Vegfb and Pdgfd-Pdgfr decreased in AAA. Whether the phenomena were the secondary changes of AAA still required further research.
In summary, our study revealed the cellular heterogeneity of Ang II-induced AAA in mice, according to sc-RNA sequencing. These ndings complement our understanding of the pathophysiological characteristics of AAA. However, there were some limitations in our study. 1) As one of the important components of the vascular media, the amount of captured vascular smooth muscle cells was insu cient. The quality inspection of single-cell RNA sequencing veri ed that our samples were up to standard. We suspected that discrepancies were likely associated with technical aspects, including the enzymatic cocktails used for single-cell release and the different approaches to data analysis. A larger sample size might eliminate the differences. 2) The aim of this study was to complete the cell map of Ang-II mediated AAA. In-depth analysis of the potential mechanisms and prognostic effects of speci c type of cell on disease were not performed in this research. In the future, we will conduct further mechanism research and functional veri cation on the interaction between Fbs and smooth muscle cells.
3) In this study we did not measure blood pressure of mice after Ang II pumping. Although previous research reported that Ang II infusion promoted AAA independent of increased blood pressure [63], it reduced the preciseness of our research.

Conclusion
In general, our study mapped the cell atlas of Ang II-induced AAA. By using sc-RNA sequencing technology, we revealed the pivotal role of broblast heterogeneity in Ang II-induced AAA. It mainly participated in the process of AAA by regulating the immune system and the metabolic balance of ECM.
Our study may provide a new perspective on emergence and development of aortic disease under different condition. Future studies should address on clinical e cacy of potential therapeutic targets selected by sc-RNA seq and validation of mechanism.